Computational efficiency of the models used in optical proximity correction (OPC) continues to be important in photolithography employing resolution enhancement. Improved hardware and model efficiency can mitigate the increase in run time resulting from the rise in modeling complexity necessary to meet advanced semiconductor process requirements. The current approach for fast image calculations in OPC utilizes the Hopkins formulation of the imaging equations combined with the sum-of-coherent-systems (SOCS) approximation. In this paper, approaches to identify the SOCS terms, which can be neglected in an OPC model applied to a given layout, are explored. One approach is quantified in terms of computational efficiency and model accuracy. The pattern-specific computational efficiency improvement ranges from 7% to 55% for six different patterning conditions used in 193 nm immersion lithography at the 45 nm semiconductor technology node.